The following guidelines should be adhered to while building queries to access CDC tables:
The application should not use Read Uncommitted or no Lock table access as a transaction isolation level when reading CDC data as it results in inconsistent or incorrect data due to lock conflict errors.
To avoid lock conflict errors, the application should restrict its data access to a time interval or accept a lock conflict as an approximate end-of-data indicator despite some records not being read. When a lock conflict occurs, some Result Set rows (determined by Fetch Array Size) are not returned to the application.
ODBC and JDBC clients have APIs to define the Fetch Array Size used to send records to the application. The default Fetch Array Size in OpenEdge SQL is 50 rows for both Drivers and for ESQL.
If the user is using the Authorized Data Truncation(ADT) feature or the Autonomus Schema Update(ASU) feature, any applications that work with data returned from SQL queries, will not get complete data. Progress suggests that users should set ADT and ASU to OFF while working with CDC tables.
The application should not use Read Uncommitted or no Lock table access as a transaction isolation level when reading CDC data as it results in inconsistent or incorrect data due to lock conflict errors.